On 4/9/21 12:09 PM, Tom Lane wrote: > One could make an argument, therefore, for holding off 0003 until > there's more support for execution-time error cursors. I don't > think we should though, for two reasons: > 1. It'd be better to keep the pg_proc representation of new-style > SQL functions stable across versions. > 2. Storing the CREATE text means we'll capture comments associated > with the function text, which is something that at least some > people will complain about the loss of. Admittedly we have no way > to re-integrate the comments into the de-parsed body, but some > folks might be satisfied with grabbing the prosrc text. >
+many for storing query text. cheers andrew -- Andrew Dunstan EDB: https://www.enterprisedb.com